The weekend is upon us, just a few more hours.
Allyson is off at another event. This time as a vendor, selling her cookbooks, and having fun.
The blue-haired fairie is with her, they are going to have a grand time.
On the nerd side of things, I am in the processing of ditching AWS. Amazon Web Services is a gateway drug. You can get started for just a few cents per month, then you suddenly find yourself shelling out serious coin.
For me, the big cost was backups. When I was at University, I was introduced to “comprehensive backups”. The idea being that we should be able to recover a file for a very, very, very long time.
When most services describe backups, they talk about having “daily” backups. That means you have 24 hours to realize that something is wrong, and recover it.
My backups are different. If you have created a file, and it is on disk when backups run, it will be backed up.
Monday through Saturday go into the daily backups. These backups are preserved for 30+ days.
Sunday backups are different, on the first Sunday of the month, backups go to the monthly backups. These are kept for 2+ years.
Finally, all other Sunday backups go on to the weekly backups. These are kept for 6 months.
So, comprehensive.
All of that takes disk space. I was exceeding 10 terabytes of data on AWS. It was running me excess dollars.
No more. I now have a local ceph cluster with multiple nodes on multiple switches, in multiple power circuits. There are over 70 Terabytes in the cluster. Two hosts have room for another 12 TB each. Two hosts can be upgraded from 12 TB to 48 TB, each. There are spots for four more hosts, each of which can handle 48 TB each.
Backups are now going to my ceph cluster. This is very redundant. I can take an entire host out of the cluster and the cluster still functions. At some point, I will configure the cluster so that I can take out multiple hosts at one time and still be fully functional.
Sorry, this is about feedback, not nerd babble.
The site is still not stable enough. The work I’m doing with ceph locally, will transfer to ceph on remote systems. Tuning ceph on the remote systems and moving the ceph cluster, proper, out of K8S will improve things greatly.
What feedback do you have for us?
Comments
2 responses to “Friday Feedback”
We the People need places like this to keep us somewhat sane…. just don’t start posting a bunch of “karen” videos like all the rest of the “conservative” sites are doing… I don’t follow any of them.. and having fun is what its all about..
Any chance of adding (signed) cookbooks to the online store?
And who on earth is the blue-haired fairie?